1234117 玖、發明說明： 【發明所屬之技術領域】 本發明係關於光學滑鼠的技術領域’尤指一種適用 光學滑鼠之移動向量動態搜尋方法及裝置。 【先前技術】 10 15 一般習知之光學滑鼠，係以一發光二極體對滑鼠墊或 其他物體進行光照射，再由一影像感測器擷取滑鼠墊上的 影像’比較前後影像而求出該光學滑鼠之移動向量，該移 動向量即為該光學滑鼠在取樣時間内所移動的距離。其運 作原理如圖1所示，其中，圓圈A表示該發光二極體照射 在滑鼠墊或其他物體時，該影像感測器所擷取之影像範 圍為進行影像比對，故取正方形B的範圍進行前後影像 匕對而為5兒明方便，三角形C係表示所擷取之影像範 圍中的一個物體。 心二：：滑鼠往左上角移動時，相對地，該影像感測 ° 衫像，則如圖1所示呈現往右下角# + 為時間tl的IS游及生- 々Γ用移動’ 不 ^ 原先所肺之料。“為時間t2 的圖形係该光學滑鼠 丁 令，標示為時間t2的圖^从 後所擁取之影像，其 ’間t2的圖形中的虛 三角形，而以實線表示—角以為原先之 技術係將該_t2時的^為移動後之三角形。習知 進行…時兩張影=:像^ 移動向量D即為該光學 ’出移動向量D。該 在取樣時間内所移動的距離。 20 1234117 然而此種習知之技術，需封整個正方 p 費許多計算時間及計算能量，因此 仃比對’需花 距離計算方法的設計仍有諸多 °先學滑鼠之移動 夕缺失而有予以改進之必要。 5 10 15 【發明内容】 本發明之目的係在提供一種光學滑鼠 態搜尋方法及裝置，以避免習知 多動向I動 , %兄為知技術因需對整個 行進行搜尋比對，而花費許多計算時 ： 題，同時，可有效降低晶體之功率消耗，而 里的問 依據本發明之-特色，係提出—種光學滑鼠之的向 ==方：’包括步驟:㈧操取一張影像.，並於此影 叹疋厂目標區域；（B)在次一取樣時間擁取下—張影 象將刖張衫像之該目標區域與此目前之影像比對， 俾於此目前之影像中找出與該目標區域相符之區域，而決 定光學滑鼠之#動向量及位移量；以及（c)再次於下一取樣 =間綠-張影像’且依據先前之移動向量而在此目前之 疋出預測區域，以將该目標區域僅與此預.測區域 比對’俾於此預測區域中找出與該目標區域相符之部分， 而更新光學滑鼠之移動向量及位移量。 依據本發明之另一特色，係提出一種光學滑鼠之移動 向里動悲搜哥裝置，主要包含一影像感測器、一記憶體裝 置及一處理單元。該影像感測器用以擷取一張影像；該記 憶體裳置用以儲存該影像感測器所擷取之影像；該處理單 凡其搞合至該記憶體，其中，當該影像感測器擷取第一張 20 1234117 ΐ:::存，憶體時’該處理單元於此影像中設定-憶體時二!該影像感測器擁取第二張影像而儲存於該記 俾於此曰Γ处理單元將該目標區域與此目前之影像比對， 定光a、、M j之〜像中找出與該目標區域相符之區域，而決 另移動向量及位移量；而當該影像感測器擷取 的移動彡像⑽存於該記憶料，處理單元依據先前 目伊區在此目前之影像中^出―預測區域，以將該 該目^ 預測區域比對，俾於此預測區域中找出與 ίο 15 移量Γ區域相符之部分，而更新光學滑鼠之移動向量及位 由於本發明設計新穎，能提供產業上利用，且確有增 進功效，故依法申請發明專利。 【實施方式】 圖2顯示本發明光學滑鼠之移動向量動態搜尋裝 置的方塊圖’其包含—影像感測器11G、-記憶體裝置100、 及y影像處理及移動向量處理單元刚。該影像感測器ιι〇 用以擷取工作平面的影像。該記憶體裝置_具有一操取影 像：憶體120、一參考影像記憶體13〇、一動態預測位移暫 存器150、及-比對範圍暫存器16()，#中，該擷取影像記 憶體120係耦合至該影像感測器11〇，用以儲存自該影像感 測器no所取擷取之現行影像。該參考影像記憶體13〇用以 儲存該擷取影像記憶體120中的先前的影像，以做為與現行 影像比對用之比對影像，該動態預測位移暫存器15〇用以儲 20 1234117 存-動態預測移動向量。該比對範圍暫存器⑽用 ί以二目標區域的範圍。該影像處理及移動向量處理 姆當鄉像❹⑶时摘取新的影像時，將 取影像記憶體12〇中原有的影像儲在 ° 5 10 15 ηΛ . ^ ㈣^㈣存至該參考影像記憶體 中，並將新的影像儲存至該擷取影像記憶體咖中 執仃影像處理及計算移動向量。 、 圖3顯示本發明之光學滑鼠之移動向量動態搜尋方法 之/爪轾圖，併請配合圖4〜圖6 意圖，於圖4中，目圈入表:：—之影像的示 取之m 影像感測器110所操 象輕圍。於此影像範圍中，以正方形 :來進行前後影像之比對。而三角 的一個物體，虛線正方形e表示該： 對乾圍暫存器140所指定之目標區域的範圍。 器η=ΓΓ左上角移動時，相對地，該影像感測 示為時間to广像則如圖4所示呈現往右下爲移動，標 干為時門1圖形表不在初始時第一次所擷取之影像，標 :=二的圖形係表示光學滑鼠往左上角移動後賴 ==广旦時間口的圖形係該光學滑鼠再往左上角 U:之影像，其中，標示為時間t2的圖形中的虛 表示原先之三角形’而以實線三角形C，表示 滑机移動後之三角形。 於圖3之流程中，首券 鼠之擷取影像記憶體120、炎考;^S201巾m學/月 位移暫存器刚及比對範圍暫^像記憶體i3G、動態預測 圍暫存态160初始化。之後，該影 20 1234117 像感測器110擷取第一張影像（時間t0)，並將該影像健存至 該擷取影像記憶體120(步驟S202)。該影像處理及移動向量 處理單元140依據該比對範圍暫存器16〇中的目標區域之範 圍’而在該所擷取的影像B”中設定一個次影像E”以作為一 5目標區域（步驟S203)，於本實施例中，該目標區域係較佳 地位於擷取之影像的中央部分。 之後’滑鼠移動而該影像感測器110再擷取一張新的 影像（步驟S204)(時間tl)，此時，該擷取影像記憶體12〇中 的先則影像將儲存至該參考影像記憶體13 〇中，而將新的影 10像儲存至該擷取影像記憶體120中，如此安排影像資料的儲 存位置係方便後續之影像比對。 於步驟S205中，影像處理及移動向量處理單元14〇將 4 一張影像之目標區域E”與目前之影像b比對，而在此目 前之影像B中找出與該目標區域E”相符之區域，藉此而決 15疋光學滑鼠之移動向量及位移量，並將該移動向量儲存於 該動態預測位移暫存器丨50(步驟S2〇6)，前述比對方法可用 習知的二階動量法（second order momentum method)，或平 均均方根誤差法（mean squared err〇r，MSE)或其他類似方 法。 20 之後，滑鼠再移動而該影像感測器11〇再擷取一張新 的影像（步驟S207)(時間t2)，此時，該擷取影像記憶體12〇 中的先前影像將儲存至該參考影像記憶體13〇中，而將新的 影像儲存至該擷取影像記憶體12〇中，該影像處理及移動向 量處理單元140依據該動態預測位移暫存器15〇中的移動向 1234117 量及該比對範圍暫存器160中的目標區域之範圍，而預估在 t2時，該目標區域的範圍會落在E，處，故由該擷取影像記 憶體120中取出一大於或等於目標區域範圍大小之現行次 影像P’作為一預測區域（步驟S2〇8)，並將該參考影像記憶 5體13()中的目標區域E僅與此預測區域p，進行比對（步驟 S209)，而在此預測區域p，中找出與該目標區域£相符之區 域，藉此而更新光學滑鼠之移動向量及位移量，並將該移 動向量儲存於該動態預測位移暫存器15〇(步驟S21〇)。前述 比對方法可用習知的二階動量法（sec〇nd 咖_她 10 meth〇d)，或平均均方根誤差法（mean squared error，MSE) 或其他類似方法。 、右在預測區域p，中無法找出與目標區域£相符之部 分，則表示該動態預測位移暫存器150中的移動向量不準 確，對於先前影像之目標區域E中所包含的三角形C，無法 15在現行影像之預測區域p，找中到相對應之三角形c，，可能 如圖5所示’該相對應之三角形C，已超出預測區域P，的範 \。故執行步驟S2G5，將目標區域赠整張現行影像B， 進行比對以找出相符之部分，並再重新計算該移動向量， ，將之回存至該動態預測位移暫存器150中。而若在整張現 20订景，像B’中無法找到與目標區域£相符之部分、其可能如 圖6所不之情形’表示先前影像之目標區域E中所包含的三 角形C，無法在現行影像B，找到相對應之三角形c，，此時 則重回步驟S2(H，並執行初始化。 1234117 由上述說明可知，本發明係以動態預測位移暫存器及 比對範圍暫存器來預測滑鼠之移動向量，依此動態預測之 移動向量來調整參考影像的選取位置及下次的影像比對範 圍以進行動態搜尋，因為無需對所擷取到影像的全畫面來 5進行搜尋，因此運算量將大幅降低。可避免習知技術需對 整個全畫面進行進行搜尋比對而花費許多計算時間及計算 能量的問題，更可有效降低晶體之功率消耗，而達省電之 目的〇 上述實施例僅係為了方便說明而舉例而已，本發明所 10主張之權利範圍自應以申請專利範圍所述為準，而非僅限 於上述實施例。 15 【圖式簡單說明】 圖1係習知之光學滑鼠運作原理之示意圖 圖4〜圖6係本發明之光學滑鼠之移動 作原理之示意圖。 向量動態搜尋方法運 【圖號說明】 影像感測器 參考影像記憶體 梅取影像記憶體 12〇 〜像處理及軸向量處理單元14〇 20 1234117 動態預測位移暫存器 記憶體裝置 150 比對範圍暫存器 160 1001234117 (1) Description of the invention: [Technical field to which the invention belongs] The present invention relates to the technical field of optical mice ', and particularly to a method and a device for dynamically searching a motion vector of an optical mouse. [Prior art] 10 15 A conventionally known optical mouse is a light emitting diode that irradiates the mouse pad or other objects with light, and then an image sensor captures the images on the mouse pad to compare the images before and after. The movement vector of the optical mouse is obtained, and the movement vector is the distance moved by the optical mouse within the sampling time. The operation principle is shown in Figure 1. Among them, circle A indicates that when the light-emitting diode is irradiated on a mouse pad or other objects, the image range captured by the image sensor is used for image comparison, so square B is taken. The range of the front and back images is 5 pairs for convenience, and the triangle C represents an object in the captured image range. Heart 2: When the mouse moves to the upper left corner, relatively, the image senses the ° shirt image, as shown in Figure 1, it appears to the lower right corner # + is the IS tour and life of time tl-々Γ used to move 'No ^ Original The material of the lungs. "The graphic for time t2 is the optical mouse Ding Ling, the image labeled as time t2 ^ the image captured from the back, the virtual triangle in the graphic between t2, and the solid line-the angle is the original The technology is to use the ^ at _t2 as the triangle after the movement. The two shadows when the training proceeds are: image ^ The movement vector D is the optical 'out movement vector D. The distance moved during the sampling time. 20 1234117 However, this conventional technique requires a lot of calculation time and energy to seal the entire square p. Therefore, the design of the calculation method of the distance to be spent still has a lot of degrees. The movement of the learner mouse is missing and improved. 5 10 15 [Summary of the invention] The purpose of the present invention is to provide an optical mouse state search method and device to avoid the conventional multiple movements. The technology requires searching and comparing the entire row. And it takes a lot of calculations: at the same time, it can effectively reduce the power consumption of the crystal, and according to the features of the present invention, it proposes-the direction of an optical mouse = = side: 'Including steps: One image. , And sigh the target area of the factory here; (B) Captured at the next sampling time-Zhang Yingxiang compares the target area of the Zhang shirt image with the current image, and clicks in the current image Find the area that matches the target area, and determine the #motion vector and displacement amount of the optical mouse; and (c) again in the next sample = green-sheet image 'and based on the previous motion vector, Identify the prediction area to compare the target area with this prediction area only. Find the part that matches the target area in this prediction area, and update the movement vector and displacement of the optical mouse. Another feature of the invention is to propose an optical mouse for searching brothers inwardly, which mainly includes an image sensor, a memory device and a processing unit. The image sensor is used to capture an image The memory is used to store the image captured by the image sensor; the processing unit is used to combine the image with the memory, and when the image sensor captures the first 20 1234117 ΐ :: : Storing, recalling the body'The processing unit in this video Set-memory time two! The image sensor captures the second image and stores it in the record. Here, the processing unit compares the target area with the current image. ~ Find the area that matches the target area in the image, and determine the motion vector and displacement. When the moving image captured by the image sensor is stored in the memory, the processing unit will In the current image ^ prediction area, in order to compare the prediction area of the target ^, in this prediction area, find the part that matches the Γ 15 movement amount Γ area, and update the movement vector of the optical mouse and Because the present invention has a novel design, can provide industrial use, and has indeed improved efficacy, it is applied for an invention patent in accordance with the law. [Embodiment] FIG. 2 shows a block diagram of a motion vector dynamic search device for an optical mouse of the present invention, which includes— The image sensor 11G, the memory device 100, and the image processing and motion vector processing unit are just now. The image sensor is used to capture an image of the work plane. The memory device has an operation image: memory 120, a reference image memory 13, a dynamic prediction displacement register 150, and a comparison range register 16 (), #. The image memory 120 is coupled to the image sensor 11 and is configured to store the current image captured from the image sensor no. The reference image memory 13 is used to store the previous image in the captured image memory 120 as a comparison image for comparison with the current image, and the dynamic prediction displacement register 15 is used to store 20 1234117 Store-dynamically predict motion vectors. The comparison range register uses the range of the two target areas. This image processing and motion vector processing is used to extract new images when the local image ❹⑶, the original images in the image memory 120 are stored in ° 5 10 15 ηΛ. ^ ㈣ ^ ㈣ is stored in the reference image memory And save the new image to the captured image memory to perform image processing and calculate the motion vector. Figure 3 shows the / claw diagram of the motion vector dynamic search method of the optical mouse of the present invention, and please refer to Figures 4 to 6 for the intent. In Figure 4, the eye circles are entered into the table ::- m The image sensor 110 operates lightly. In this image range, use square: to compare the front and back images. For an object in a triangle, a dotted square e indicates the range of the target area specified by the register 140. When the device η = ΓΓ moves in the upper left corner, the image sensing is shown as time to wide image, as shown in Figure 4, showing movement to the lower right, and the standard mark is that the time gate 1 graphic table is not the first time The captured image, the symbol marked with == 2 indicates that the optical mouse is moved to the upper left corner, and the == Guangdan time port is the image of the optical mouse further moved to the upper left corner with U :, which is marked as time t2 The imaginary in the figure represents the original triangle 'and the solid triangle C represents the triangle after the slider moves. In the process of Fig. 3, the first mouse captures image memory 120 and inflammation test; ^ S201 学 m / month shift register just now and comparison range temporary ^ image memory i3G, dynamic prediction surrounding temporary state 160 initialization. After that, the image 20 1234117 image sensor 110 captures the first image (time t0), and stores the image in the captured image memory 120 (step S202). The image processing and motion vector processing unit 140 sets a secondary image E ”in the captured image B” as a 5 target area according to the range of the target area in the comparison range register 160. Step S203). In this embodiment, the target area is preferably located at the center of the captured image. After 'the mouse moves and the image sensor 110 captures a new image (step S204) (time t1), at this time, the previous image in the captured image memory 12 will be stored to the reference The image memory 13 is stored in the new image 10 and the new image 10 is stored in the captured image memory 120. The arrangement of the image data storage location is convenient for subsequent image comparison. In step S205, the image processing and motion vector processing unit 14 compares the target area E "of the four images with the current image b, and finds the current image B that matches the target area E" Area to determine the motion vector and displacement of the 15 疋 optical mouse, and store the motion vector in the dynamic predictive displacement register 50 (step S206). The aforementioned comparison method can be used in the conventional second order Momentum method (second order momentum method), or mean squared error method (MSE) or other similar methods. After 20, the mouse moves again and the image sensor 11 captures a new image (step S207) (time t2). At this time, the previous image in the captured image memory 12 will be saved to The reference image memory 13 is stored, and a new image is stored in the captured image memory 120. The image processing and motion vector processing unit 140 moves to 1234117 according to the movement in the dynamic prediction displacement register 15. And the range of the target area in the comparison range register 160, and it is estimated that at t2, the range of the target area will fall at E, so a greater than or The current secondary image P 'equal to the size of the target region is used as a prediction region (step S208), and the target region E in the reference image memory 5 (13) is compared with this prediction region p only (step S209), and in the prediction area p, find an area that matches the target area £, thereby updating the movement vector and displacement of the optical mouse, and storing the movement vector in the dynamic prediction displacement register 150 (step S21). The aforementioned comparison method may use a conventional second-order momentum method (second_10 meth〇d), or a mean squared error (MSE) method or other similar methods. If you ca n’t find a part in the prediction area p that matches the target area £, it means that the motion vector in the dynamic prediction displacement register 150 is inaccurate. For the triangle C contained in the target area E of the previous image, It is impossible to find the corresponding triangle c in the prediction area p of the current image. As shown in FIG. 5, 'the corresponding triangle C has exceeded the range of the prediction area P'. Therefore, step S2G5 is performed, the target area is presented with the entire current image B, and a comparison is performed to find a matching part, and the motion vector is recalculated, and stored in the dynamic prediction displacement register 150. And if the scene is ordered in the whole 20, like B 'can not find the part that matches the target area £, it may be as shown in Figure 6' indicates that the triangle C included in the target area E of the previous image cannot be found in In the current image B, the corresponding triangle c is found. At this time, the process returns to step S2 (H and performs initialization. 1234117 As can be seen from the above description, the present invention uses a dynamic prediction displacement register and a comparison range register. Predict the motion vector of the mouse, and adjust the selected position of the reference image and the next image comparison range for dynamic search based on this dynamically predicted motion vector, because there is no need to search the full screen of the captured image. Therefore, the calculation amount will be greatly reduced. It can avoid the problem that the conventional technology needs to search and compare the entire full screen and spend a lot of calculation time and calculation energy. It can also effectively reduce the power consumption of the crystal and achieve the purpose of power saving. The embodiments are merely examples for the convenience of description. The scope of the rights claimed in the present invention should be based on the scope of the patent application, rather than being limited to the above. 15 [Schematic description] Figure 1 is a schematic diagram of the operating principle of a conventional optical mouse. Figures 4 to 6 are schematic diagrams of the moving principle of the optical mouse of the present invention. Vector dynamic search method operation [illustration of drawing number] ] The image sensor refers to the image memory and takes the image memory 12〇 ~ image processing and axis vector processing unit 1420 1234117 dynamic prediction displacement register memory device 150 comparison range register 160 100
1212